Income Tax Progressivity and Nonreligion in Central and Eastern Europe: A Case of the Czech Republic

Abstract

Our paper focuses on the tax progressivity and nonreligion in central and eastern Europe using an example of the Czech Republic, one of the most atheistic countries in the world. Religion might imply formal affiliation with a certain confession or active church attendance, but it might also become a determinant of taxation preferences. We employ ordinal regression analyses to study direct and mediation effects of both church affiliation and church attendance on a representative sample from the Czech Republic (n = 1924, 54.8% female, aged 18–95, M ± SD: 52.0 ± 16.9; 19.4% with higher education) controlling for employment status, social class and socio-demographics. The results suggest that neither church affiliation nor church attendance were related to desired income tax progressivity; social class plaid the most important role. The frequency of church attendance was significantly related to the perceived adequacy of taxation of higher incomes, where the more respondents attended the religious services, the more they believed that the taxes on the rich are too high. However, the churches’ ideas to help the needy were manifested in the preferences for international tax progressivity, where the frequent churchgoers were more inclined to the idea that the rich countries should pay additional taxes to help the poor countries. These controversial results may indicate the rivalrous position of the church and the state in the Czech Republic in terms of taxation of the wealthy. We suggest that under the condition of no church tax, the state taxation of the rich may be viewed as diverting funds, which could otherwise be directed to the church. These results might be of some interest to the state, the church and to the academic researchers alike and significantly contribute to the discussion on specific features of nonreligiosity in central and eastern Europe.

1. Introduction

While the numbers of nonreligious individuals (i.e., those who do not associate themselves with any religion) is growing worldwide, little attention has been paid in the research literature on the determinants of nonreligion in central and eastern Europe, a region marked with a turbulent history, the forty-year period of communist regimes and, most recently, the aftermath of the economic transformation and institutional changes (Bognár and Kmetty 2020; Bubík et al. 2020; or Trzebiatowska 2021).

Our research is based upon the assumption that one of the determinants of religion and religious behavior in central and eastern Europe might be income taxation as a proxy of trust to the state and reliance upon the state, which was demonstrated by a study from the Czech Republic by Strielkowski and Čábelková (2015). Our paper deals with the specific features of the relationship between church attendance as a mediator of religion and preferences for progressive income taxation. The paper builds upon previous research dealing with the research of religion and other social factors, for example the link between religion, science and secularization (Pinelli and Einstein 2019), or religion and political economy (McCleary and Barro 2006), religion and international migration (Strielkowski et al. 2016), religion and depression (Anderson et al. 2021), religion and sexual upbringing (Krull et al. 2021) or church attendance and religious changes in a traditional society (Vezzoni and Biolcati-Rinaldi 2015; Kortt et al. 2017). In addition, it is also based upon research focused on the relationship between tax avoidance and religion (Jun and Yoon 2018; McGee et al. 2020; or Cho and Yoon 2020). We also consider issues such as the influence of religiosity on the civic engagement (Fałkowski and Kurek 2019), the link between religiosity and economic growth (Hirschle 2010; Hirschle 2013) or simply the general concept of the relationship between taxes and religionism based on religion as such (Teather 2006).

Unlike many other studies that focus on the study of religion and other social factors (Gouveia et al. 2020), our research deals with the issue of the relationship between religion and preferences for progressive income taxation in the Czech Republic. This is a relationship between religion and economic but also social preferences. As Loury (2004) or Stegmueller (2013) point out, relatively little work identifies the influence of religion on such economic issues as taxation and willingness to pay taxes, even though this issue has a long tradition, especially in terms of the link between social classes and religion (Schwadel et al. 2009). Thence, it appears to be important to distinguish preferences for progressive income taxation according to social classes too.

In our research, we use the church attendance indicator as a measure of religionism, as is the case in other research that confirms the influence of church attendance on the researched target factor (see Loury 2004). However, this may not always be the case for all target factors (McKenzie 2001). The fact that church attendance does not have to be positively correlated with wages in all social groups or the level of income, measured by the log of GDP per capita, has been discussed in the literature (see Mohanty 2020; or Herzer and Strulik 2017). In addition, Lewis (2002) found no significant association between the frequency of church attendance and happiness. All of the above confirms inconsistent results in the study of religion and socio-economic factors. Hence, there is plenty of room for examining this area, which has its own specifics.

Therefore, in our research, we focus on determining whether tax preferences are related to the frequency of church visits and arrive at the following hypotheses and goals for our research. This approach is based on Dyreng et al. (2012) who demonstrate that religion might influence economic choices and outcomes in quite a number of contexts.

In our view, there is a concept of social identity (the respondent’s affiliation with a religious group, a church) linked to the economic preferences, as stated in another context by Costa-Font and Cowell (2015), which might prevail in central and eastern European countries, such as the Czech Republic. Social identity has become accepted as a key concept on which the endogeneity of economic behavior and preferences are based. Similarly, Goeke-Morey et al. (2015) showed, in a case study from Northern Ireland, that social identity can be “multifaceted, with historical, religious, political, social, economic, and psychological underpinnings”.

The aim of this paper is to test the association between tax preferences and religion. The Czech Republic was chosen because it is a country where religion and the attitude toward the church have been marked by several facts and facts that do not occur elsewhere (outside central and eastern Europe) (see Barro and McCleary 2003, p. 36).

This paper is structured as follows. Section 2 describes the materials and methods and features the research hypotheses tested in this research. Section 3 describes the data used for our empirical model. Section 4 Contemplates over the determinants of religion and church attendance in the Czech Republic. Section 5 presents the empirical models employed in our paper. Section 6 outlines the main results of the model and provides the discussion of these results. Finally, Section 7 features overall conclusions with policy implications and pathways for further research.

2. Materials and Methods

For many decades, religiosity was negatively associated with government regulation of the religion market and with the religious oppression that accompanied the presence of a communist government in central and eastern Europe (Gautier 1997). The elimination of communist regimes at the end of the 1990s and beginning of 1990s led to a recovery of religiosity in most central and eastern European countries (Evans and Northmore-Ball 2012). However, one has to note that there are some differences between the CEE countries with varying approach to the rediscovery of God and the church (Radu 2013, p. 1231). With regard to this, Savage (2020, p. 108) concludes that religiosity has a fundamentally different effect on the preferences for redistribution of wealth in CEE countries compared to Western democracies—in CEE countries, people tend to favor state intervention for increasing this redistribution.

In the Czech Republic, the Catholic religion is the most widespread, even though the majority of residents do not profess any religion at all (in fact, over 70% claim to be atheists (see Furstova et al. 2021). There are issues such as returning the land and real estate to the Catholic church or no particular taxes being imposed on the church. On some occasions, the church might be even viewed as a financial rival to the state (especially when it comes to the restitution of land and real estate).

Similar to other CEE countries, individual philanthropy in the Czech Republic does not have a long tradition (Čábelková et al. 2015) as, for example, in the United States (see Kim 2013). The empirical section will find out whether the more people go to church, the more they think that taxes on the rich are too high and, therefore, they cannot pay as much to the church. Historically, the wealthy people were the biggest contributors of the Catholic church, similarly as the wealthy were supposed to help the poor in all kinds of charity. The contributions of high income to the state in the form of progressive taxation may be viewed as rivalrous to the church members, as it diverts the funds that could otherwise be directed to the church. However, one has to acknowledge that there can also exist a positive relationship, e.g., due to the fact that religious people tend to be in favor of more redistribution from the rich to the poor or are, in general, suspicious toward wealth accumulation (see, e.g., Keister 2003; Koudelková et al. 2015; or Berggren and Bjørnskov 2013). It might be the existence of economic inequality and this distortion of people’s sympathies that allows and perhaps encourages the rich to neglect the most elementary norms of moral behavior (Khoshnava et al. 2019). There is a strong correlation between inequality and religion, such that societies characterized by high inequality are more religious than societies with a more egalitarian income distribution. For example, race provides much better support for wealth than income or any other variable. In addition, in many countries nowadays, the poor may not need massive redistribution because they think social mobility can make them rich tomorrow.

In the Czech Republic, there is no religious tax collected by the state for the purposes of the churches (this kind of tax exists, for example, in Germany). The churches in the Czech Republic are financed from the contributions provided by the state, which are too small to cover all the expenses, the voluntary contributions of locals, the entrance fees to some of the most famous religious buildings for tourists and from its own production. The churches, especially the Catholic church, have also benefitted from restitution of property originally confiscated by the communists.

In order to achieve our aims and goals, we formulated the following research hypotheses based on the previous literature review and the analysis of the available sources:

Hypothesis H1.

Tax preferences are related to the religion of the respondent.

Hypothesis H2.

Tax preferences are related to the frequency of visits to the church.

Hypothesis H3.

The frequency of visits to the church mediates the association between the religion of the respondent and tax preferences.

3. Data

The data were collected by the Czech Institute of Sociology in the international project International Social Survey Programme in 2019 (Sociologický ústav Akademie věd ČR 2019). One thousand nine hundred twenty-four respondents from the Czech Republic were asked to fill out the questionnaire voluntarily and anonymously (n = 1924, 54.8% female, aged 18–95, M ± SD: 52.0 ± 16.9; 19.4% with higher education). The response rate was 50.3%. The selection of the respondents relied on a three-level stratified probabilistic selection using the personal (and non the household-level) data. The samples were designed to be representative adult citizens and adults of any nationality residing in the Czech Republic and aged 15 and over. The register of census districts and buildings of the Czech Statistical Office from the spring of 2018 and variables from the 2011 census were used to obtain information concerning the complete list of buildings located in the Czech Republic and the number of persons living in them (Sociologický ústav Akademie věd ČR 2019). The selection of respondents was carried out with respect to their location within provinces (NUTS levels), cities and towns, as well as the economic status that corresponded to the average values recorded for the Czech Republic. The respondents included in the sample were previously recruited for the research data panel to be carried out by the Czech Institute of Sociology over a longer period of time, either personally using a probabilistic sampling and via telephone or during personal interviews.

The data are available for non-commercial use upon signing up the corresponding contracts with the depositor of the data (International Social Survey Programme) and should not be transmitted to third parties. They can be obtained from the International Social Survey Programme upon written request.

When it comes to the choice of indicators used in our paper, church attendance was chosen as a factor of religion on economic preferences due to the fact that church attendance has already proved itself in other researches, such as Loury (2004, p. 125), Schwadel et al. (2009), Ingen and Moor (2015), Sawkins et al. (1997), Vezzoni and Biolcati-Rinaldi (2015), Paleček (2017) or Berggren and Ljunge (2021), who also used church attendance in their research as a measure of religiosity. Three indicators of tax preferences are considered (see

Table 1. Indicators of tax preferences: the distributions of the respondents (in %).

Indicator	Answers, %
How much should high earners pay low earners in taxes? 1	Much higher percentage	Higher percentage	The same percentage	Smaller percentage	Much smaller percentage	Undecided 4
	10.10	44.00	38.90	3.10	1.00	2.90
Taxes on high earners are: 2	Too high	High	Adequate	Low	Too low	Undecided 4
	2.50	9.80	41.70	33.30	6.80	5.90
Rich countries should pay poor countries 3	Strongly agree	Rather agree	Neither agree nor disagree	Rather disagree	Strongly disagree	Undecided 4
	9.20	25.60	28.00	21.40	13.40	2.50

¹ Do you think that high-income people should pay a higher, equal or lower percentage of income to low-income people in taxes? (1) Much higher percentage, (2) Higher percentage, (3) The same percentage as low-income people, (4) Smaller percentage, (5) Much smaller percentage, (6) Undecided. ² In your opinion, what are the taxes on people with high incomes today? (1) Too high, (2) High, (3) Adequate, (4) Low, (5) Too low, (6) Undecided. ³ People in rich countries should pay an extra tax as a contribution to help people in poor countries. (1) Strongly agree, (2) Rather agree, (3) Neither agree nor disagree, (4) Rather disagree, (5) Strongly disagree, (6) Undecided. ⁴ Undecided respondents were excluded from the data analysis.

). In Table 1, the responses represent the % of respondents answering each of the three questions (first column)—the exact wording of the question is presented below the table in the footnotes.

It also needs to be noted that the “high-income people” and “low-income people” mentioned in Table 1 were the categories with certain levels of income according to the Czech national statistics and legal definitions applied by the Czech National Bank and the Ministry of Finance.

4. Determinants of Religion and Church Attendance in the Czech Republic

We approach the distribution of respondents according to religion because different results may occur, as described in Jones et al. (2019, p. 409): “Christians are found to be more pro-market than those who are not religious.” Similarly, Benk et al. (2015, p. 2012) came to the conclusion that the attitude toward cheating on taxes (committing a tax fraud) might differ by religion.

The sole fact of professing a religion should not necessarily suffice for changing one’s values and behavioral patterns. Participation in religious services seems to be a better indicator of practicing religion as a form of transmission of group values projected by religion. Our paper employs both the indicators of belonging to a particular religious group and the frequency of participation in services (see

Table 2. Indicators of religion and attendance or religion services: the distribution of the respondents (in %).

Indicator	Answers, %
Religion 1	Roman Catholic	Evangelical	Hussites	Other Christian	Other non-Christian	None	Missing
	23.60	2.10	1.20	0.90	1.00	68.50	5.50
Church attendance 2	Never	Less than once a year	Once or twice a year	Several times a year	Once a month and more often	No answer/missing
	59.30	11.20	9.70	7.80	10.20	1.80

¹ Do you profess religion and, if so, what religion do you profess? ² If we do not count exceptional events, such as funerals or weddings, how often do you attend religious services? (1) Never, (2) Less than once a year, (3) About once or twice a year, (4) Several times a year, (5) About once a month, (6) Two to three times a month, (7) Almost every week, (8) Every week, (9) Several times a week, (10) I do not know, no answer. Due to the low number of observations, the answers were aggregated as shown in the table.

). Interestingly, the percentage of respondents who never attend religious service is somewhat smaller in the sample than the proportion of respondents denying belonging to some religious group (59.30% and 68.50%, respectively; see Table 2, which follows). This comparison among non-religious people and “church attenders” becomes more significant if one considers that the respective question was about church attendance that excluded exceptional events.

Given the extremely imbalanced representation of religions provided in Table 2, the variable religion was transformed to a binary variable equal to 1 for the respondents denying belonging to any religious denomination (68.50% of the respondents, see Table 2) and being equal to 0 for the rest. Respondents whose answers were missing were excluded from the analysis.

Expectedly, the respondents professing some religion visited religious services more often. The relevant means and 95% confidence intervals are presented in Figure 1.

Table 3. Socio-demographic indicators and the distribution of the respondents (in %).

Indicator	Answers, %
Income (difficulty to make ends meet) 1	Very difficult	Rather difficult	Not easy or difficult	Rather easy	Very easy	Undecided
	3.80	15.60	42.70	29.30	7.10	1.40
Social class 2	Lower class	Working class	Lower middle class	Middle class	Higher middle class	Higher class	Undecided
	4.10	18.70	21.30	44.00	8.50	1.60	1.80

¹ If we take into account the total income of your household, which includes all sources of income of all members who contribute, how difficult is it for your household to make ends meet? ² Most people classify themselves as a social group (class). Which social group (class) would you classify yourself in? We also control for employment status (Employee or entrepreneur, Unemployed, Student, Apprentice, Retired disabled, Retired, Housewife (home carer), On maternity or parental leave, Other), year of birth, education, gender and emotional condition (the subjective level of happiness at the moment, 10-point scale).

that follows presents an overview of the socio-demographic indicators and the distribution of the respondents.

5. The Model

In our empirical model presented hereinafter in this section, we rely upon a set of ordinal regression analyses to study both the direct effect presented in the hypotheses H1 and H2 (Formulas (1) and (2)) and the mediation effects of church attendance on the association between the religion of the respondents and tax preferences (Formula (3)).

We employ the approach described in McCleary and Barro (2006, p. 49) who used religion as an independent variable affecting individual characteristics, such as work ethic, honesty and thrift and, as a result, economic performance.

Tax preferences = logit (a₀ + a₁Religion + a₂EmploymentStatus + a₃SocialClass + a₄YearOfBirth +
a₅Education + a₆Gender + a₇Income + a₈EmotionalCondition + e₁)

(1)

Tax preferences = logit (a₀ + a₁ChurchAttendance + a₂EmploymentStatus + a₃SocialClass +
a₄YearOfBirth + a₅Education + a₆Gender + a₇Income + a₈EmotionalCondition + e₂)

(2)

Tax preferences = logit (a₀ + a₁ChurchAttendance + a₂Religion + a₃EmploymentStatus +
a₄SocialClass + a₅YearOfBirth + a₆Education + a₇Gender + a₈Income +
a₉EmotionalCondition + e₃)

(3)

where

• TaxPreferences—three indicators of national and international tax progressivity as presented in Table 1, consequently;
• Religion—dummy equals to 1 if the respondent does not profess any religion and 0 otherwise (see Table 2);
• ChurchAttendance—the frequency with which the respondents attend religious services (see Table 2);
• EmploymentStatus—set of dummies employee or entrepreneur, unemployed, student, apprentice, retired disabled, retired, housewife (home carer), on maternity or parental leave, other. The last category was excluded to prevent singularity;
• SocialClass—social class (Table 3);
• YearOfBirth, Education, Gender—year of birth, education, gender;
• Income—a proxy of income (difficulty to make ends meet, see Table 3);
• EmotionalCondition—the subjective level of happiness at the moment (10-point scale).

6. Results and Discussion

In this section, we present the results of the empirical model, which was defined in the previous sections, and discuss the results stemming from this model. The results of ordinal regression analyses (Formulas (1)–(3)) are presented in

Table 4. Religion is related to tax preferences. The results of ordinal regression analysis (Formula (1)).

	Desired Tax Progressivity		Tax Adequacy for High Incomes		International Tax Progressivity
	Estimate	Sig.	Estimate	Sig.	Estimate	Sig.
Threshold = 1	−2.885 **	0.001	−4.931 ***	0.000	−0.924	0.257
Threshold = 2	−0.346	0.693	−3.085 ***	0.000	0.813	0.318
Threshold = 3	2.679 **	0.002	−0.763	0.371	2.016 *	0.013
Threshold = 4	4.400 ***	0.000	1.606	0.061	3.272 ***	0.000
Religion	−0.060	0.567	−0.082	0.429	−0.207 *	0.033
Employment status
Employee or entrepreneur	0.469	0.493	−0.768	0.252	0.830	0.198
Unemployed	0.084	0.911	−0.713	0.329	−0.041	0.953
Student	0.846	0.257	−0.839	0.251	0.752	0.283
Apprentice	1.436	0.231	−1.596	0.168	−0.752	0.502
Retired disabled	0.303	0.678	−0.657	0.359	1.050	0.128
Retired	−0.010	0.988	−0.268	0.691	0.759	0.242
Housewife (home carer)	1.015	0.252	−0.794	0.388	0.556	0.509
On maternity or parental leave	0.415	0.567	−0.533	0.455	0.670	0.326
Social class
Lower class	−2.041 ***	0.000	1.277 **	0.005	−0.304	0.480
Working class	−1.708 ***	0.000	0.802 *	0.038	−0.298	0.416
Lower middle class	−1.270 **	0.002	0.454	0.232	−0.264	0.464
Middle class	−1.227 **	0.002	0.370	0.318	−0.044	0.899
Upper middle class	−0.924 *	0.028	0.103	0.793	−0.208	0.576
Socio-demographics
Year of birth	0.000	0.079	0.000	0.099	0.000 *	0.011
Education (years)	0.012	0.510	−0.028	0.130	0.038 *	0.025
Gender (man)	−0.253 **	0.008	0.405 ***	0.000	−0.050	0.579
Income (difficulty to make ends meet)	0.158 **	0.004	−0.088	0.112	0.220 ***	0.000
Emotional condition (Happiness)	0.001	0.969	−0.035	0.214	−0.015	0.556
Pseudo R-Square
Cox and Snell	0.067		0.066		0.043
Nagelkerke	0.076		0.072		0.045
McFadden	0.032		0.028		0.014
Sig.	0.000		0.000		0.000
n	1740		1687		1740

Reference variables: gender—women, employment category—other, social class—higher, no religion. Link function: Logit. ***—significant on 0.1% level. **—significant on 1% level, *—significant on 5% level.

,

Table 5. Church attendance as related to tax preferences. The results of ordinal regression analysis (Formula (2)).

	Desired Tax Progressivity		Tax Adequacy for High Incomes		International Tax Progressivity
	Estimate	Sig.	Estimate	Sig.	Estimate	Sig.
Threshold = 1	−2.048 *	0.029	−4.791 ***	0.000	−1.659	0.056
Threshold = 2	0.499	0.594	−3.017 **	0.001	0.092	0.916
Threshold = 3	3.506 ***	0.000	−0.702	0.441	1.322	0.128
Threshold = 4	5.240 ***	0.000	1.677	0.066	2.583 **	0.003
Church Attendance	−0.022	0.509	−0.154 ***	0.000	−0.144 ***	0.000
Employment status
Employee or entrepreneur	1.063	0.147	−0.253	0.724	0.508	0.459
Unemployed	0.691	0.383	−0.203	0.792	−0.506	0.495
Student	1.474	0.062	−0.318	0.680	0.408	0.579
Apprentice	2.054	0.094	−1.069	0.366	−1.033	0.367
Retired disabled	0.959	0.215	−0.246	0.745	0.650	0.370
Retired	0.573	0.436	0.262	0.716	0.398	0.563
Housewife (home carer)	1.613	0.080	−0.190	0.842	0.236	0.788
On maternity or parental leave	1.011	0.189	−0.011	0.988	0.295	0.682
Social class
Lower class	−1.754 ***	0.000	1.187 **	0.009	−0.371	0.388
Working class	−1.420 **	0.001	0.721	0.061	−0.374	0.307
Lower middle class	−0.953 *	0.018	0.350	0.356	−0.333	0.355
Middle class	−0.929 *	0.018	0.270	0.465	−0.118	0.737
Upper middle class	−0.597	0.152	−0.028	0.943	−0.257	0.490
Socio-demographics
Gender (man)	−0.222 *	0.020	0.373 ***	0.000	−0.058	0.513
Income (difficulty to make ends meet)	0.161 **	0.004	−0.097	0.077	0.200 ***	0.000
Year of birth	0.000	0.108	0.000	0.143	0.000 *	0.010
Education (years)	0.011	0.538	−0.038 *	0.040	0.041 *	0.017
Emotional condition (Happiness)	−0.006	0.817	−0.026	0.364	−0.020	0.435
Pseudo R-Square
Cox and Snell	0.066		0.078		0.054
Nagelkerke	0.074		0.085		0.057
McFadden	0.031		0.033		0.018
Sig.	0.000		0.000		0.000
N	1756		1706		1758

Reference variables: gender—women, employment category—other, social class—higher, no religion. Link function: Logit. ***—significant on 0.1% level, **—significant on 1% level, *—significant on 5% level.

and

Table 6. Church attendance and religious affiliation as related to tax preferences. The results of ordinal regression analysis (Formula (3)).

	Desired Tax Progressivity		Tax Adequacy for High Incomes		International Tax Progressivity
	Estimate	Sig.	Estimate	Sig.	Estimate	Sig.
Threshold = 1	−2.170 *	0.022	−4.685 ***	0.000	−1.722 *	0.049
Threshold = 2	0.383	0.686	−2.838 **	0.002	0.028	0.974
Threshold = 3	3.408 ***	0.000	−0.480	0.602	1.252	0.152
Threshold = 4	5.194 ***	0.000	1.916 *	0.038	2.509 **	0.004
Church Attendance	−0.017	0.679	−0.221 ***	0.000	−0.173 ***	0.000
Religion	−0.044	0.736	0.337 *	0.011	0.147	0.229
Employment status
Employee or entrepreneur	1.063	0.147	−0.236	0.742	0.504	0.462
Unemployed	0.772	0.331	−0.237	0.760	−0.432	0.561
Student	1.443	0.068	−0.282	0.716	0.430	0.560
Apprentice	2.046	0.096	−0.969	0.414	−1.004	0.381
Retired disabled	0.895	0.248	−0.173	0.819	0.698	0.337
Retired	0.587	0.425	0.235	0.744	0.378	0.583
Housewife (home carer)	1.621	0.079	−0.266	0.780	0.225	0.797
On parental leave	1.017	0.188	−0.033	0.965	0.326	0.652
Social class
Lower class	−1.908 ***	0.000	1.346 **	0.004	−0.491	0.262
Working class	−1.560 ***	0.000	0.878 *	0.026	−0.463	0.215
Lower middle class	−1.107 **	0.008	0.543	0.162	−0.412	0.262
Middle class	−1.074 **	0.008	0.404	0.286	−0.227	0.526
Upper middle class	−0.739	0.084	0.146	0.716	−0.384	0.310
Socio-demographics
Gender (man)	−0.244 *	0.011	0.387 ***	0.000	−0.054	0.552
Income (financial difficulty)	0.164 **	0.003	−0.101	0.071	0.195 ***	0.000
Year of birth	0.000	0.104	0.000	0.121	0.000 **	0.008
Education (years)	0.012	0.522	−0.030	0.107	0.040 *	0.020
Happiness	−0.003	0.912	−0.025	0.389	−0.010	0.714
Pseudo R-Square
Cox and Snell	0.068		0.082		0.054
Nagelkerke	0.076		0.089		0.056
McFadden	0.032		0.034		0.018
Sig.	0.000		0.000		0.000
n	1729		1677		1729

Reference variables: gender—women, employment category—other, social class—higher, no religion. Link function: Logit. ***—significant on 0.1% level, **—significant on 1% level, *—significant on 5% level.

that follow below. Ordinal logistic regression is a statistical analysis method that can be used to model the relationship between an ordinal response variable and one or more explanatory variables. The thresholds 1 to 4 reported in Table 4 are the measures for each possible increase in the outcome variable.

The threshold coefficients (from 1 to 4) in Table 4 are not interpreted individually (which is typical for interpreting and analyzing the results of the ordinal regression analysis). These thresholds only represent the intercepts, specifically the point (in terms of a logit) where the tax preferences might be predicted into different categories.

From Table 4, it follows that people professing some religions tend to agree that rich and wealthy countries should pay an additional tax to help poor countries. In order to shed more light upon the results, let us explain how they should be interpreted. For example, in the column “Desired tax progressivity”, one can see that the “lower class” has a coefficient −2.041, and it comes through as highly significant (at 0.1% level). This means that a change in the predictor variable makes the event more likely. In accordance with the logic of the logit model that is employed and computed in this paper (and widely described in the econometric literature that can be consulted for those readers interested in technical specifics, which are not closely related to the scope of this very journal), mostly the signs of the estimated coefficients are important (hence the “-“ sign in this particular case should be noted), but the magnitude (the value of the numbers) also reveals some information. Positive signs indicate that, ceteris paribus, the odds of approving tax progressivity are rising with an increase in the explanatory variable. If one is interested in the magnitude, taking the value of exp of the estimate yields the odds ratio, since coefficients are actually log odds ratios. This interpretation also helps to understand the difference in the attitude toward desired tax progressivity between, e.g., lower class and working class.

To be more specific and reader friendly, we can further comment on the results for different social classes in the “Desired tax progressivity” column. Scrolling further down the column, one can see that the “working class” has a coefficient −1.708, which also comes through as highly significant (at 0.1% level) as in the case of the “lower class”. Thence, the odds of approving tax progressivity are decreasing with an increase in the explanatory variable but with a lower magnitude than in the case of the “lower class”. Further down the column, the coefficients for the “lower middle class”, “middle class” and “upper middle class” are also negative, but the magnitudes are lower (−1.270, −1.227 and −0.924, respectively). It also has to be noted that the significance was lowered to the 1% (“lower middle class” and “middle class”) and 5% (“upper middle class”) levels. These results indicate that all the social classes, when compared to the higher social class (“upper middle class”), tend to prefer higher tax progressivity. In the case of the two remaining models (see the columns “Tax adequacy for high incomes” and “International tax progressivity”), the results for all social classes came through as insignificant (except for the “lower class” estimate in model 2.

These findings are in accord with the results obtained by Mohdali and Pope (2014, p. 71) who found that religiosity had a minimal but statistically significant positive impact on voluntary tax compliance. Social identity thus becomes crucial in decision making, since according to Costa-Font and Cowell (2015), empirical evidence shows that social identity carries weight in explaining the presence of social preferences and attitudes toward redistributive institutions.

Furthermore, we find that men prefer higher tax progressivity than women and tend to believe more (somewhat populistic and often used thesis) that taxes for high-income people are rather low. It becomes apparent that the easier it is for the respondents to make ends meet, the lower the preferred tax progressivity is, both on the national and international levels.

Additionally, we found that the more years of education the respondents have, the less they believe that rich countries should pay an additional tax to help poor countries.

Yet, another interesting finding comes from estimating the age factor. Our results show that the older the respondents are, the more they believe that rich countries should pay an additional tax to help the poor ones. Here, we obtain similar results as Quattrociocchi (2018, p. 1415) who found that members of groups with higher mean incomes tend to have weaker preferences for redistribution, all else being equal. In addition, the following outcomes can be derived from Table 4:

• People who profess some religions are more in agreement that rich countries should pay an additional tax to help poor countries;
• All social classes prefer higher tax progressivity compared to the higher social class;
• Men prefer higher tax progressivity than women and believe that taxes for high-income people are rather low;
• The easier it is for respondents to make money, the lower the preferred tax progressivity is at the national and international levels;
• The older the respondents are, the less they believe that rich countries should help poor countries;
• The older the respondents are, the more they believe that rich countries should pay an additional tax to help poor people in general;

Additionally, from the results reported in Table 5, it follows that the more often people attend religious services, the more they believe that the taxes for the rich are too high.

Here, we are in accord with Reda (2010, p. 312) who tested the relationship between religious and economic preferences with the argued propositions that individuals who have high levels of religiosity and, as such, belong to conservative denominations, will prefer to give less to the government in the form of taxes. Furthermore, it appears that the more often people attend religious services, the more they agree that rich countries should pay additional taxes to help poor countries. While this result might seem contradictory, it can be explained as such: when a person stresses her or his unconformity with the taxes within her or his country, such a person does not express absolutely a “pro-market” position. Instead, she or he only thinks that there are objectively too many taxes. If this same person is in favor of raising taxes to help the poor countries, it means that she or he thinks that her or his own country is rich enough and prefers to give her or his money to other ends. Thence, the problem is not the taxation system as such but the use of the money and the distribution of wealth.

Similarly, our results show that all the social classes, compared to higher social class, prefer higher tax progressivity. We also found that men prefer higher tax progressivity than women, and more believe that taxes for high-income people are rather low. It turned out that the easier it is for the respondents to make ends meet, the lower the preferred tax progressivity is, both on the national and international levels.

Regarding the level of education and age of our respondents, it turned out that the more years of education the respondents had, the less they believed that rich countries should pay an additional tax to help poor countries. Furthermore, the older the respondents were, the more they believed that rich countries should pay an additional tax to help the poor ones.

From Table 6 it follows that the more often people attend religious services, the more they believe that the taxes for the rich are too high. It also turns out that the more often people attend religious services, the more they agree that rich countries should pay additional taxes to help the poor countries.

• All the social classes compared to higher social class prefer higher tax progressivity;
• Men prefer higher tax progressivity than women, and more believe that taxes for high-income people are rather low;
• The easier it is for the respondents to make ends meet, the lower the preferred tax progressivity is, both on the national and international levels;
• The more years of education the respondents have, the less they believe that rich countries should pay an additional tax to help poor countries;
• The older the respondents are, the more they believe that rich countries should pay an additional tax to help the poor ones.

The new finding is the mediating effect of visits to religious ceremonies on the association between tax preferences and religion. The visit to religious ceremonies is the co-founder of the association between religion and international tax progressivity. The associations of visits to religious ceremonies and tax adequacy of high income are different for people professing a codified religion and not. In both cases, visits to religious ceremonies seem to be more important for forming tax preferences than the declaration of belonging to some religious group.

7. Conclusions

Overall, our findings show that tax and religiosity preferences have not yet been properly studied, either internationally or in the case of central and eastern European countries. Recent research on tax preferences has rather focused on the relationship between tax preferences and political party affiliation (Roemer 1998) or on the issue of religiosity and economic inequality (see Solt et al. 2011; or Highfill and O’Brien 2019). However, some research has shown that religiosity can have a strong overlap in many realms of the human society, including the economic and social ones, as demonstrated by Iannaccone (1998), McKenzie (2001) or Hoffmann (2013).

Armed with these knowledge and findings, in this paper, we studied the association of tax preferences and respondents’ religion using a sample from the Czech Republic, a post-communist country in central and eastern Europe known for its high level of atheism and nonreligiosity. Two proxies of religion were chosen: the proclaimed belonging to a certain religious denomination and the active participation in religious ceremonies. Our results suggest that the preferred tax progressivity is not related to the religion of the respondent but is rather influenced by her or his subjective social class. This result is rather surprising, as the Christian types of religion (especially Roman Catholic Christianity, which attracted most of the respondents) tend to declare help to the needy. Here, our results were somewhat in disaccord with Stegmueller (2013, p. 1073) who postulated that religion shaped up the economic preferences of individuals. In a figurative sense, our findings can be used in further research to examine religion and tax morality in the Czech Republic (or any other CEE country), so there may be other findings than those reported by Torgler (2006) or Strielkowski and Čábelková (2015) who claimed that there was strong evidence for the influence of religiosity on the tax morale.

Additionally, from our study in the Czech Republic, we found that the more the respondents visit religious ceremonies, the more they tend to believe that the taxes imposed on the rich people are too high. This result is also surprising, as one would expect that religion preaches help for the needy by the rich and wealthy. Even though one hardly thinks about the economic situation of her or his church when asked about taxes, this result can perhaps be explained by the specifics of the relationship between the state and the church in the Czech Republic when many people perceive the church as under-financed by the state and needing more money that is consumed in the forms of the taxes. The only model where the hypothesis that religion teaches the rich and wealthy to support the poor was confirmed was the case of international tax progressivity, meaning that the populations of rich countries should be additionally taxed to help the poor countries. This idea of taxing the rich countries in order to pay the poor countries was positively related to the frequency of attendance of religious ceremonies. This attendance was more important than the proclaimed belongingness to some religious denomination. The result of this hypothesis is, therefore, in accordance with a Tkáčová and Slivka (2021) who showed that the Christian ethics perspective is based on the biblical command to care for and protect creation.

We can relate this single confirmed conclusion to the fact that when deciding on economic preferences, church attendees identify themselves with their economic social status rather than with their social status in the form of religion, so we do not contradict the Costa-Font and Cowell conclusions (Costa-Font and Cowell 2015, p. 357) that social identity carries weight in explaining the presence of social preferences and attitudes toward redistributive institutions. However, some findings are too general, as Kirchmaier et al. (2018, p. 282) found that religious people yield lower preference for redistribution.

Therefore, we dealt with the specific aspects of nonreligion in the Czech Republic and the influence of church attendance as a mediator of religion on the preferences for progressive income taxation. Further research on this interesting topic might address the influence of other economic factors on religiosity in central and eastern Europe and post-communist countries. For example, it would be interesting to perform a cross-country comparison of all so-called “Visegrad group” countries (Czech Republic, Hungary, Poland and Slovakia) or compare the outcomes separately for the transition economies and the developed Western countries. Furthermore, it would be interesting to perform a comparison with the case of Russia where the communist rule lasted thirty years longer, and the position of the church was more deeply marginalized.

As for the pathways for further research on the topic of nonreligion in central and eastern Europe in general, and in the Czech Republic in particular, one interesting issue is the relationship between the determinants of tax morale, as suggested by Doerrenberg and Peichl (2013). Overall, we can evaluate any research in the field of economics and religion as beneficial (see Iyer 2016). In light of our findings, we can agree with Hoffmann (2013, p. 29) who claims that while religion is important for the development, culture, as well as the political and social discourse, it functions as one of alternative social group identifiers. In addition, another interesting pathway for future research might be the interdependence of religiosity or nonreligiosity and the institutional variables, such as property rights and the rule of law. With regard to this, a cross-country regression analysis, including the institutional outcome variables, might be very useful and would likely produce valuable results. Furthermore, it appears to be interesting to use the interaction term of church attendance and religion and also include the interaction with the covariates. Such analysis, comprising all central and eastern European countries, would shed more light on the issues of nonreligion and the social and cultural determinants of economic institutions, but it would also require very precise and comparable data, such as those we obtained for the Czech Republic, which might be costly and cumbersome to collect.

All in all, we show that nonreligion in CEE countries might be projected through the economic and financial aspects that reflect the clash between the state and the church, as well as the redistribution of wealth that followed the economic transformation and transition.